ppu-paddle-ocr/processor/paddle-ocr.service.d.ts

Lightweight, probably the fastest PaddleOCR SDK in TypeScript. Runs anywhere JavaScript runs: Node.js, Bun, Deno, mobile react-native, web browsers, and browser extensions. Docker & CLI supported. The official SDK is browser-only. Accurate text detection

import type { Canvas } from "ppu-ocv";
import { BasePaddleOcrService } from "../core/base-paddle-ocr.service.js";
import type { PaddleOptions, RecognizeOptions } from "../interface.js";
import type { FlattenedPaddleOcrResult, PaddleOcrResult } from "../web/paddle-ocr.service.web.js";
/**
 * PaddleOcrService - Provides OCR functionality using PaddleOCR models.
 * To use this service, create an instance and call the `initialize()` method.
 */
export declare class PaddleOcrService extends BasePaddleOcrService {
    /**
     * Creates an instance of PaddleOcrService.
     * @param options - Configuration options for the service.
     */
    constructor(options?: PaddleOptions);
    /**
     * Fetches a resource from a URL and caches it locally.
     * If the resource is already in the cache, it loads it from there.
     */
    private _fetchAndCache;
    /**
     * Loads a resource from a buffer, a file path, a URL, or a default URL.
     */
    private _loadResource;
    /**
     * Not used by the Node service — initialization runs in {@link initialize}.
     * @throws Always; call {@link initialize} instead.
     */
    protected initSessions(): Promise<void>;
    /**
     * Initializes the OCR service by loading models and dictionary.
     * This method must be called before any OCR operations.
     */
    initialize(): Promise<void>;
    /**
     * Checks if the service has been initialized with models loaded.
     */
    isInitialized(): boolean;
    /**
     * Changes the detection model for the current instance.
     * @param model - The new detection model as a path, URL, or ArrayBuffer.
     */
    changeDetectionModel(model: ArrayBuffer | string): Promise<void>;
    /**
     * Changes the recognition model for the current instance.
     * @param model - The new recognition model as a path, URL, or ArrayBuffer.
     */
    changeRecognitionModel(model: ArrayBuffer | string): Promise<void>;
    /**
     * Changes the text dictionary for the current instance.
     * @param dictionary - The new dictionary as a path, URL, ArrayBuffer, or string content.
     */
    changeTextDictionary(dictionary: ArrayBuffer | string): Promise<void>;
    /**
     * Recognize text in an image, returning flattened results.
     *
     * @param image - Source image as `ArrayBuffer` or `Canvas`.
     * @param options - Must include `flatten: true`.
     * @returns Flat list of recognized text items.
     */
    recognize(image: ArrayBuffer | Canvas, options: RecognizeOptions & {
        flatten: true;
    }): Promise<FlattenedPaddleOcrResult>;
    /**
     * Recognize text in an image, returning line-grouped results.
     *
     * @param image - Source image as `ArrayBuffer` or `Canvas`.
     * @param options - Optional recognition options.
     * @returns OCR results grouped by detected text lines.
     */
    recognize(image: ArrayBuffer | Canvas, options?: RecognizeOptions & {
        flatten?: false;
    }): Promise<PaddleOcrResult>;
    /**
     * Pre-download all default models to the local cache directory.
     * Call this before `initialize()` to avoid first-run download delays or
     * to warm the cache in CI/CD pipelines and test suites.
     */
    static downloadModels(options?: {
        verbose?: boolean;
    }): Promise<void>;
    /**
     * Delete all locally cached ONNX model files.
     */
    clearModelCache(): void;
    /**
     * Release all ONNX sessions and free resources.
     */
    destroy(): Promise<void>;
}
export default PaddleOcrService;